The purpose of this paper is to consider how to make use of qualitative data in educational research, focusing on the description of the socialization process, by examining the theoretical and empirical relationships between T. Parsons' theory of socialization, R.F. Bales' observation system of social interaction, and H. Sacks' perspective of conversation analysis. Parsons' theory of socialization consists of several theoretical concepts including pattern variables, aspects of social control, and aspects of deviant behavior. These concepts have a close connection with Bales' system of twelve categories for making observations of types of social action, covering a wide range of behavior in everyday life. Although Bales' categorization system was originally devised for performing quantitative analysis in experimental situations, it is of use in data description in qualitative research. In other words, Parsons' socialization theory, regardless of whether it is qualitative or quantitative, has much to do with the direct observation of social action. Bales thought that we could understand social actions right then and there. In contrast, Sacks had a clear idea that it should take a long time for us to describe even small pieces of observational data. In common with Bales, however, he also placed a high value on the careful classification of types of social action. This paper thus attempts to ascertain how these two ways of observation play a complementary role in the study of socialization by analyzing the same data for mother-child interactions from both points of view. In qualitative educational research, it is important to continue to examine the details of each case, but it is also necessary to be conscious of the relationship between data analysis and the theoretical perspective in order to maintain the validity of data description.